Vietnamese Swimming's Missing Split Data: The Cost of a One-Line Results Sheet

Open the results page of an international swim meet and I get twelve data columns for a single swim: reaction time to the hundredth of a second, 50m split, 100m split, average velocity, stroke rate, turn time at every wall. Open the results page of most domestic meets in Vietnam and I get four columns: name, year of birth, club, final time. The distance between those two pages is not on the pool deck. It sits in the data layer beneath the pool deck.

I still keep one such file on my machine. Full names, full clubs, empty everywhere else. An empty record. For an analyst, a file like that has no usable value. It tells me nothing beyond how many people got into the water.

For a decade, Vietnamese swimming lived in the reflected light of a single name. Nguyen Thi Anh Vien, born in 2026 in Can Tho, competed at three consecutive Olympic Games — London 2026, Rio 2026, Tokyo 2026 — and brought home a collection of SEA Games medals plus podium finishes at continental level. She deserved every one of them. But there is a consequence few people mention: the presence of one outstanding athlete conceals questions about the system underneath. When you already have someone swimming fast, nobody is forced to explain why everyone else swims slowly.

When Anh Vien left the elite stage, the questions surfaced immediately: who is next, in which event, on what improvement trajectory. Those are questions most swimming nations in the region can answer with data. We cannot, because we never collected the right things to answer them.

The cause is not laziness. It is that this sport, in Vietnam, still runs on a single metric: the final time. Over many years sitting in the stands at domestic meets, I noticed a repeating habit. When a swimmer touches the wall, the reporter asks for the time. The coach reads out the time. The crowd remembers the time. Nobody asks how the two hundred metres were actually distributed.

A 200m freestyle race in a 50m pool consists of four lengths and three turns. If all I know is the final result, I know exactly one thing: whether that swimmer was faster or slower than the person beside them. I do not know whether the start was good, how much the three turns cost, whether the first half was faster than the second, or what the stroke rate was. Those four questions lead to four different training programmes. A single number cannot separate them.

A more concrete example. Elite reaction times at the start typically fall between 0.60 and 0.75 seconds, measured from the signal to the moment the toes leave the block. In the 50m freestyle, the gap between a 0.62 reaction and a 0.78 reaction is 0.16 seconds — roughly 1.5% of a swim completed in under 22 seconds. In the shortest event, 1.5% is enough to change the podium. But if the results sheet has no reaction column, a coach cannot know whether to invest more time in the starting block or in the swimming itself. They will choose by feel, and feel at the starting block is usually wrong, because reaction time is measured in hundredths while the human eye resolves about a tenth.

Now the splits. A 400m individual medley swimmer can divide the race in two opposite ways: fast first half and a fading second, or a cautious first half and a surging finish. Same final result, two different physiological profiles. The first says the aerobic base is not yet thick enough. The second speaks to pacing control and confidence in the finish. The training programme for these two athletes cannot be the same. But if the federation's archive only stores the final time, both walk into the next training season with identical data: one number.

Then comes a deeper layer — what swimming calls efficiency metrics. Two swimmers can both complete the 100m freestyle in 52 seconds with entirely different structures. The first reaches 52 seconds through high stroke rate and short distance per stroke. The second reaches 52 seconds through low stroke rate and long distance per stroke. From the outside, both are talents. From the data, these are two opposite development paths. The first still has room to extend distance per stroke, meaning they can get faster still by building pulling strength. The second is close to their ceiling, and the only way forward is to raise stroke rate — a high-risk technical change that easily disrupts breathing rhythm.

Once again I have to return to my three-source habit. At international meets, I cross-check the organiser's results against the official timing provider's data and against the world federation's records. Three sources, three confirmations, and only then do I write. With domestic swimming in Vietnam, that habit hits a structural obstacle: there is no second source. The organiser's published result is the only source. No independent timing system, no third party recording splits, no open database for cross-checking. A single number standing alone, and a single number standing alone cannot be verified.

I once treated models as scripture. Now a model is a compass — but without it, you are lost.

In football, I once gave a name to an invisible pressure that every team fears: Binh Duong pressing. In swimming, that invisible pressure carries a different name. It lives in the fifteen metres after every turn.

Swimming rules limit the underwater phase. In freestyle, backstroke and butterfly, a swimmer must surface within the first fifteen metres after leaving the wall or after the start. In breaststroke, only one dolphin kick is permitted after the start and after each turn. Those rules exist because the underwater phase is faster than the surface phase — less turbulence, lower drag. Which means every wall creates a technical window that can win or lose tenths of a second.

A swimmer losing 0.3 seconds per turn against regional standard loses nearly 1 second over 200m freestyle and more than 2 seconds over 400m individual medley. Set against the margin between a medal and a place in the final at regional level, that is not a small gap. Yet it does not appear on any results sheet. To measure it, someone must time each wall separately and record it in a table, consistently, every swim, every season.

This is where the absurdity comes in. In football I learned that xG is not wrong — the sport itself is irrational, so since 2026 I have counted the irrational part into my models. Swimming has its own irrationality; the difference is that it sits beneath the surface, in exactly the layer that every timing model treats as a constant.

I am talking about the data layer of the pool itself. Pool depth directly affects wave drag. A three-metre-deep pool dissipates the waves created by swimmers far better than a two-metre pool, which is the minimum competition rules allow. The same swimmer at the same fitness level will produce two different numbers in those two pools. Lane rope type, water temperature, the quality of the filtration system — all of them feed into the final number.

So when a domestic results sheet records a national record, I need three more facts before I will read it: how long the pool is, how deep it is, and whether it was measured by automatic timing or by hand. Without those three facts, the number is still there, but its weight has changed.

And this is where numbers become slippery. A 25m short-course pool produces more turns than a 50m long-course pool. Every turn is a push-off faster than average swimming speed. Which means the same swimmer, at the same effort, will always post a faster time in a 25m pool than in a 50m pool. Converting between the two with a fixed formula is a fantasy, because the gain at each turn depends on the individual, the event and the moment.

In Vietnam, the number of 50m competition pools can be counted on one hand. Most junior meets and many national-level meets take place in 25m pools. That is a physical constraint, not anyone's fault. But when a national ranking mixes short-course times with long-course times and has no pool-code column, that ranking produces something more dangerous than missing data: data that is wrong while appearing valid.

Numbers do not lie, but people always find a way to lie with them.

Here I have to say something that runs against the usual instinct of the analytics crowd. The default response to missing data is to go and find data. I think that reflex is wrong, at least at this stage.

Vietnam's swimming problem has never been a shortage of data. We generate data every time there is a meet. What we lack is a data-reading layer — a group of people, a process, and a habit of archiving. If tomorrow someone installed per-wall timing at every domestic meet, we would have a mountain of numbers and still be unable to answer a single question, because nobody is trained to read them and there is no storage standard for comparing across seasons.

By contrast, a modest reading layer could start immediately with no large budget. One extra timekeeper at each wall, one fixed form recording three columns: split, reaction, pool code. Sustain that long enough and those three columns generate something more valuable than any prediction model — the improvement trajectory of each individual athlete over time. Once you have trajectories, you begin to see who will hit a ceiling and who still has room to travel.

I once built a prediction model from a large European football dataset and called most of the knockout matches in Russia in 2026. What produced the results was not the model. What produced the results was that I already had data for the model to learn from. When the field is empty, every model collapses. I rebuild from the burnt remains of data.

Vietnamese swimming sits in exactly that state: a pool deck full of talent, a pile of half-burnt data, and a next generation growing up with very little information about itself. What is worth noting is that we know precisely what is missing. Not a grand discovery — just three data columns left blank for twenty years.

Reputation is only a name. What remains is always how you read the contest.

If you run a swim team, the practical problem is not how to acquire the most modern measurement system. It is this: at the next meet, who will sit at the end of the pool and record three numbers nobody has ever recorded. That person needs no qualifications. Only a sheet of paper and enough patience to do it for three consecutive seasons. A good model needs years of data, and the season begins the moment the first whistle sounds.